AI Shopping Agents Are Coming But No One Is Ready For The Challenges Ahead
AI shopping agents face significant hurdles in becoming a mainstream tool for consumers, according to industry leaders, despite growing interest in their potential. Matt Maher, founder and CEO of M7 Innovations, cited security protocols, lack of agentic commerce standards, and retailer policies as barriers preventing AI agents from completing purchases. Melissa Bridgeford, CEO of Wizard Commerce, noted that even for product discovery, AI models like OpenAI’s ChatGPT provide specific recommendations only 9% of the time.
OpenAI’s decision to abandon its Instant Checkout feature, which allowed purchases via chat, led to partnerships with retailers like Walmart dissolving. Bridgeford said the industry is moving toward enabling agentic commerce but faces unresolved issues around fraud, refunds, and returns. Courtney Robinson, head of policy at Akoya, emphasized that liability for unintended purchases remains undefined, with no standardized framework for determining responsibility.
Why Agentic Commerce Faces Regulatory and Technical Roadblocks
The absence of clear standards for AI-driven transactions has created a fragmented landscape. Maher highlighted that even if companies use terms and conditions to limit liability, “perceptual liability” could still hold merchants accountable if a customer disputes a purchase. For example, an AI agent accidentally buying a blazer from the Gap could prompt a loyal customer to demand a refund despite contractual disclaimers.
Cybersecurity concerns further complicate adoption. Norman Menz, CEO of Flare, warned that AI agents could expand the “attack surface” for fraud, as malicious actors might hijack legitimate agents or create synthetic ones using stolen data. Adam Winnick of Finality suggested blockchain technology could help verify AI agents’ identities but acknowledged the need for open-source systems to manage this process.
What May Happen Next in Agentic Commerce
Industry leaders predict a gap between consumer demand for AI shopping tools and the development of safeguards. Menz said markets may adopt AI agents before resolving fraud risks, mirroring the evolution of “card not present” transactions in e-commerce. Ben Leventhal of Blackbird Labs, which focuses on dining reservations, noted that identity verification remains a critical challenge, though payment fraud is less of a concern in the restaurant sector.
Blockchain-based solutions are seen as a potential pathway, but experts agree that standards will take years to establish. Leventhal expressed optimism, stating that “magical” AI tools will eventually overcome current limitations, though merchants may initially bear the risk of fraud.
Frequently Asked Questions
What challenges prevent AI agents from completing purchases? Security protocols, retailer policies, and the absence of agentic commerce standards are major obstacles, according to Matt Maher of M7 Innovations.
How does liability for AI purchases remain unresolved? No clear standards exist for determining responsibility if an AI agent makes an unintended purchase, as noted by Courtney Robinson of Akoya.
What role could blockchain play in AI commerce? Blockchain technology may help verify AI agents’ identities and ensure they act on behalf of legitimate users, as suggested by Adam Winnick of Finality.
As AI shopping agents evolve, how might businesses balance innovation with the need for consumer protection?